General characteristics of selected children

Among the 84 patients, 41 were boys, and 43 were girls, with age ranging from 2 months to 14 years and an average of 4.4 years. There were 18 (21.4%) children with basic diseases, 6 (7.1%) had congenital heart disease; 4 (4.8%) had bronchial asthma; and primary ciliary dyskinesia, hepatoblastic carcinoma, renal malformation, giant gyri malformation, diabetes, chronic granuloma, pelvic tumour, and congenital cleft palate was each reported in 1 child. The most common symptoms were cough (92.9%), fever (73.8%), and dyspnoea (35.7%), followed by poor mental health (27.4%), stridor (26.2%), and chest pain (11.9%), and the last was haemoptysis (2.4%). A total of 65 cases (77.4%) reported moist rales, and 26 (31%) had diminished respiratory sounds. In addition, 54 (62.3%) children exhibited severe lung imaging, including 36 with pleurisy (42.9%), 28 with inflammatory consolidation (33.3%), 24 with (28.6%), 20 with pleural effusion (23.8%), 20 with atelectasis (23.8%), 11 with necrosis (13.1%), and 7 with interstitial abnormalities (8.3%). The white blood cell (WBC) count (12.57 ± 6.46 × 109/L) was calculated, as was the neutrophil ratio (77.06 ± 11.96%), the lymphocyte ratio (13.98 ± 6.30%), the C-reactive protein (CRP) (77.13 ± 59.58 mg/L), and the procalcitonin (PCT) (11.03 ± 17.57 ng/mL). Before admission, 72 (85.7%) patients had been treated with antibiotics and 20 (23.8%) with glucocorticoids (Table 1).

Table 1 Clinical information of childrenThe pathogen detection of SCAP

The most common pathogen detected using mNGS was Streptococcus pneumoniae (25 strains), human herpesvirus type 5 (12 strains), Haemophilus influenzae (11 strains), and Mycoplasma pneumoniae (MP) (11 strains). Pathogens, such as oral Streptococcus, Bifidobacterium brevis, Bacteroides fragilis, and Candida proximalis, were also detected. Combined with clinical manifestations, the bacteria were considered as respiratory colonising bacteria, not pathogenic ones (Fig. 2).

Fig. 2

All microorganisms detected by Metagenomic Next-Generation Sequencing

The most common pathogens detected using CT were MP (20 strains), S. pneumoniae (13 strains), and Staphylococcus aureus (10 strains).

Finally, 76 cases (90.5%) were diagnosed with determinate pathogen infection, and 18 (23.7%) were bacterial infections. Six patients (7.9%) were infected with a virus, and there were 10 (13.2%) cases of atypical bacterial infection. In addition, there were three (3.9%) cases of Mycobacterium tuberculosis (MTB) infection and 39 (51.3%) of mixed infection (Fig. 3A).

Fig. 3

A Metagenomic Next-generation sequencing combined Conventional technology detected the pathogen classification. B Metagenomic Next-generation sequencing combined Conventional technology diagnose the type of polymicrobial infection

The detection rate of mNGS and CT

The sensitivity, specificity, positive predictive value and negative predictive value of mNGS were 92.1% (70/76), 62.5% (5/8), 95.9% (70/73) and 45.5% (5/11), respectively. mNGS detection rate was 83.3% (70/84), which was higher than that of the CT at 63.1% (53/84) (P = 0.003).

The bacterial detection rate of mNGS was higher than that of the CT (80.8% [42/52] vs. 44.2% [23/52], P < 0.001) (Fig. 4A). The detection rate of Gram-negative bacteria was significantly higher in mNGS than that in the CT (90.1% [20/22] vs. 22.7% [5/22], P < 0.001), while there was no significant difference in Gram-positive bacteria (75% [30/40] vs. 57.5% [23/40], P = 0.098). S. pneumoniae was the most detected bacteria in 29 cases, in which the detection rate of mNGS was higher than that of the CT (89.2% [25/29] vs. 44.8% (13/29), P = 0.001). A total of nine cases were detected using mNGS and CT, sixteen were only positive for mNGS, and four were only positive for CT. The second most detected bacteria were H. influenzae (12 cases), in which the detection rate of mNGS was higher than that of the CT (91.7% [11/12] vs. 33.3% [4/12], P < 0.005). mNGS alone detected eight cases, CT alone detected one, and common method detected three. There were 12 cases of S. aureus, three were detected using mNGS and CT, two were only positive for mNGS, and seven were positive for CT. Among the three children with Enterococcus faecalis infection, two were detected using CT and one using mNGS. E. faecium and Micrococcus gamboge were identified by culture but not by mNGS in patient 59. There were three cases of Klebsiella pneumoniae, one in patient 11, one in patient 44, detected using mNGS, and one in patient 64, detected using CT. There were two cases of Pseudomonas aeruginosa, one of which was detected using both mNGS and CT and the other using mNGS alone. Additionally, mNGS detected Prevotella melanogenes in patient 26, Enterobacter cloacae in patients 43 and 44, Escherichia coli in patient 50, and Acinetobacter baumannii in patient 65, all of which were not detected on CT.

Fig. 4

Comparison of pathogens detected by Metagenomic Next-generation sequencing (mNGS) and Conventional technology (CT). A Bateria detected by mNGS and CT methods. B Virus detected by mNGS and CT methods. C Fungus detected by mNGS and CT methods. D Atypical pathogen detected by mNGS and CT methods. ADV, adenovirus; CMV, human herpes virus; EBV, Epstein-Barr virus,; RSV, Respiratory syncytial virus

For virus detection, the detection rate of mNGS was higher than that of the CT, but there was no statistical significance (76% [19/25] vs. 56% [14/25], P = 0.136) (Fig. 4B). Respiratory syncytial virus (RSV) and Epstein-Barr virus (EBV) were the most frequently detected viruses, with seven cases each. mNGS alone identified five cases of RSV, while CT alone identified one case; however in patient 79, RSV was detected using both mNGS and CT. EBV was detected by CT in six cases (patients 9,10, 13, 53, 63, and 65), and by mNGS in four cases (patients 13, 36, 63, and 65). Boca virus (HBoV) was detected in four cases, all of which were detected by mNGS. In three cases of CMV infection, both mNGS and CT were positive. Two cases of adenovirus (ADV) and one case of influenza A virus were detected using CT. However, mNGS detected bocavirus in patients 39, 42, 68, and 76, rhinovirus in patient 58, and human metapulmonary virus in patient 66.

The detection rate of fungi using mNGS was higher than that of the CT (81.8% [9/11] vs. 18.2% [2/11], P = 0.004) (Fig. 4C). Pneumocystis vulgaris (four cases) was the most frequently detected fungus, followed by C. albicans (three cases). There was one case of Cryptococcus neoformans, Aspergillus aspergillus, and Aspergillus niger. Only two instances of C. albicans were discovered using CT in patients 62 and 83. However, mNGS missed only patient 83. In patient 26, mNGS and CT were both negative, but the serum G test (1, 3-D glucan test) was positive (301 pg/mL), which matched the clinical symptoms and imaging diagnosis of fungal infection.

There was no significant difference in the detection rate of MTB between mNGS and CT (60% [3/5] vs. 80% [4/5], P > 0.005). A total of two cases were detected using mNGS and CT, one case was detected using mNGS alone, two cases were detected using CT, three were detected using MTB-Xpert in CT, and one by culture method (Table 2).

Table 2 The detection of M. tuberculosis by mNGS and CT

Among atypical bacterial infections, mNGS detection rate was lower than that of the CT (50.0% [12/24] vs. 91.7% [22/24], P = 0.003) (Fig. 4D). CT found one case of Legionella pneumophila and one of Chlamydia trachomatis alone. The detection rate of MP using mNGS was lower than that of the CT (52.4% [11/21] vs. 95.2% [20/21], P = 0.004). In addition, 10 cases were positive for both mNGS and CT, one for mNGS alone, and 10 for CT alone. CT included MP-polymerase chain reaction (PCR) detection in 57.1% (12/21), PA detection in 52.4% (11/21), and GICT detection in 14.3% (3/21) (Table 3).

Table 3 The detection of MP by mNGS and CTThe mixed infection of SCAP

There were 39 (51.3%) cases of mixed infection. CT alone detected 16 (21.1%) cases of mixed infection, and mNGS alone detected 20 (26.3%) cases of mixed infection. However, this difference was not statistically significant (P > 0.005).

Mixed bacterial and viral infections were the most common type, with 12 (30.8%) cases. Eight (20.5%) patients had mixed bacterial and fungal infections and five (12.8%) cases of mixed bacterial and atypical pathogens (Fig. 3B).

Diagnosis consistency between mNGS and CT

A total of 49 (58.3%) cases were positive for both mNGS and CT, 22 (26.2%) for only mNGS, 5 (6.0%) for only CT, and 8 (9.5%) were negative for both CT and mNGS.

Among 49 double-positive cases, mNGS and CT results were completely consistent in 16 (19.0%) patients and completely inconsistent in 14 (16.7%) cases. Nineteen (22.6%) cases were partially consistent, indicating that at least one detected pathogen overlapped between CT and mNGS (Fig. 5).

Fig. 5

Consistency between Metagenomic Next-generation sequencing and Conventional technology diagnosis

Changes after the detection of mNGS

There were 26 (31.0%) patients who changed their treatment based on mNGS results, and their symptoms were reduced. Antibiotics were changed in nine patients, antibiotics were added in five patients, antifungal drugs in nine, antivirals in seven, and antituberculosis drugs in one (Table 4).

Table 4 Changes in treatment of children with mNGS